Uncertainty Tolerance - The Next Human Weakness
How AI’s smallest conveniences are train us out of sitting with uncertainty
For years, the pervading view has been that modern media tech has been a vehicle of distraction and influence. Television was taking over our evening and social media is colonizing the little pauses between activities and pushing us into echo chambers, and of course Google rescued us when those factoids escaped us.
These media channels clearly brought some pretty seismic changes, but even these older tools tended to leave a gap. They could interrupt thought, reward avoidance, or make knowledge feel closer than it was, but they usually stopped somewhere just outside the moment of effort. A person still had to compare, judge, draft or consider which allowed us to feel dim for a minute, and decide what they meant.
Calling AI another attention machine or the destroyer of critical thinking underplays the hand that AI is holding. It’s not that AI lures people away from the lumpy problems, but it’s the small ones, where it arrives at the moment a tiny hurdle presents itself and then generously offers to save you from this micro annoyance.
This all feels very harmless, partly because that moment was nothing more than a frivolous little annoyance, but also that low-effort, high-return solution is very good at presenting itself as common sense and efficiency.
The useful irritation we kept mistaking for waste
These old school tools like Google and even newspapers preserved more resistance than we give them credit for. A Google search required a small act of discipline: forming the question, reading across results, and learning to tell the useful page from one that was just a rambling poorly executed opinion.
Inside these infinitesimal irritations, there was often a little training effect. They invisibly kept the person in contact with uncertainty.
Most adult competence is built on these moments in the middle of a task. We’ve all taken several runs at an email, either because the tone felt wrong or that second paragraph didn’t quite do the point justice.
These frictions are unpleasant because they expose the gap between having a thought and being able to articulate it, which may be exactly why they have been easier to dismiss than to value.
AI is sand between the stones
AI does not feel like a bulldozer in its ordinary everyday use. It feels more like a natural conclusion to the point that you would have got to anyway. My analogy is that I see it a bit like sand, finding the little spaces between bigger ideas rather than completely building the entire structure.
The visible furniture of work remains in place — these bigger stones or ideas are still there, they’re just more refined and supported.
AI pours into the gaps between those stones.
It fills the cognitive, personal spaces where hesitation used to live, existing inside the doubt helping to make you feel less awkward and uncertain about a decision, phrase or tone.
Nobody builds a productivity case around these moments because they are pathetically small, inferior and difficult to quantify, but they are also where people practise staying mentally present under pressure.
AI is attractive because it makes those moments disappear without making the task disappear. The work still gets done, the email is written and the options compared and so the task reaches its inevitable close, but just a bit more efficiently.
It feels like efficiency because the visible output arrives faster but the loss is not only effort, it is the small discovery, inside the task and thus the effort the task was asking for.
So what? Surely a world without these tiny annoyances frees us up for the chunkier challenge.
The point at which AI becomes hardest to resist is usually the point at which uncertainty rises, any uncertainty, big or small and as it turns out, we have a lot of uncertainties throughout our day.
It is the compound effect of reaching for AI in all of these uncertainties that is the worrying pattern.
What begins as a sensible one-off soon becomes practical enough to repeat, and after a few repetitions the uncertain moment has acquired a reflex response.
A person feels unsure, reaches for the machine, receives immediate relief, and the task becomes easier.
At that exact moment, AI offers a clean bargain. “Need help compiling this message?” Oh go on then.
And this is the dangerous little loop. The capacity that is spared from use becomes less available next time. The next uncertain moment feels heavier because the person has become less practised at sitting inside it.
Uncertainty is also a subtle judgement on oneself. It’s that unsettling feeling of inadequacy where you have the suspicion that the next ten minutes may reveal a limit one had hoped to keep hidden from yourself.
These moments of rescue from uncertainty will breed atrophy of those tiny muscles that built a foundation of self-reliance. And soon its absence starts to feel normal.
This is not as grand as critical thinking, it’s more a case of solving ‘micro solutions’ in moments where we typically would overcome normal everyday obstacles.
The person is not choosing weakness in some grand moral sense. The person is accepting help, repeatedly, at the point where the muscle would otherwise have worked.
Minuscule moments of difficulty were doing more than slowing us down
The educational psychologists have been saying a version of this for years.
Robert and Elizabeth Bjork’s work on “desirable difficulties” argues that some conditions which make learning feel harder in the short term can improve longer-term retention and transfer. Spacing, variation and interleaving can make practice less smooth, but the roughness is part of the point. The Bjork Learning and Forgetting Lab describes spacing and interleaving as desirable difficulties that can benefit learning performance.
But I’m not saying that all struggle is good. Plenty of struggle is just bad teaching or bad software and deserves to be bypassed.
Life without AI helps us form judgment because it keeps a person with the problem long enough to notice its shape. It creates little encounters with our own limits.
AI is very good at removing the feeling of that encounter by supplying an answer before the person has properly met the question.
A recent study in Societies looked directly at AI tools, cognitive offloading and critical thinking, using survey data and interviews with 666 participants. It reported that more frequent AI tool use was associated with lower critical-thinking scores, with cognitive offloading acting as a mediating factor; the paper is suggestive rather than a final verdict, but it points at the same loop this piece is circling.
The private cost is very likely to show up as a lower tolerance for uncertainty, a weaker appetite for revision, and a more fragile relationship with blankness, until ordinary thinking starts to feel like unnecessary friction because a machine can produce something passable before the discomfort has had time to teach anything.
The danger is not that everyone is now going to become immediately incompetent. Instead, it’s that competence becomes harder to distinguish from fluent assistance as the person can still produce the artefact, but the private assembly may have gone missing.
Some friction deserves to die
The argument, then, is not to keep things hard for the sake of it. It is to preserve the ability to tell the difference between friction that wastes people and friction that forms them. Yes — this is nigh on impossible, unless it’s AI itself that recognises, regulates and resists our reach for instant resolution.
Most people and organisations will prefer not to ask the question “is this a righteous friction?”. After all, who cares if the work is getting done quicker to a decent standard.
We must all try to hold on to these micro moments — resist taking the lift when the stairs are clearly a better longer term strategy otherwise climbing will start to feel like a strange old skill from a harder time.
Thanks for reading Confidently Wrong. Subscribe for weekly arguments about AI, work and the strange things people start doing when technology makes them feel efficient.